The use of tumour-selective prodrugs (relatively inactive compounds that can be selectively converted to more active compounds in vivo) is a valuable concept in cancer therapy (see for example, Denny, Eur. J: Med. Chem. (2001) 36,577). For example a prodrug may be converted into an anti-tumour agent under the influence of an enzyme that is linkable to a monoclonal antibody that will bind to a tumour associated antigen. The combination of such a prodrug with such an enzyme/monoclonal antibody conjugate represents a very powerful clinical agent. This approach to cancer therapy, often referred to as “antibody directed enzyme prodrug therapy” (ADEPT), is disclosed in international publication WO/1988/007378.
A further therapeutic approach termed “virus-directed enzyme prodrug therapy” (VDEPT) has been proposed as a method for treating tumour cells in patients using prodrugs. Tumour cells are targeted with a viral vector carrying a gene encoding an enzyme capable of activating a prodrug. The gene may be transcriptionally regulated by tissue specific promoter or enhancer sequences. The viral vector enters tumour cells and expresses the enzyme, in order that a prodrug is converted to an active drug within the tumour cells (Huber et al., Proc. Natl. Acad. Sci. USA (1991) 88, 8039). Alternatively, non-viral methods for the delivery of genes have been used. Such methods include calcium phosphate co-precipitation, microinjection, liposomes, direct DNA uptake, and receptor mediated DNA-transfer. These are reviewed in Morgan & French, Annu. Rev. Bioehem., 1993, 62; 191. The term “GDEPT” (gene-directed enzyme prodrug therapy) is used to include both viral and non-viral delivery systems (Denny et al U.S. Pat. No. 6,310,237). One example of a non-viral delivery system being the tumour colonising bacteria Clostridia, utilised in an approach termed clostridia-directed enzyme prodrug therapy (CDEPT).
Many nitroaromatic compounds can be reduced by both mammalian and bacterial flavoprotein enzymes, which effect stepwise addition of up to six electrons. The major enzymatic metabolite is usually the 4-electron reduced species (hydroxylamine). A number of nitrophenyl mustards and nitrophenylaziridines have been reported as prodrugs for use in gene-directed enzyme prodrug therapy (GDEPT) in conjunction with nitroreductase enzymes. In particular, CB 1954 [5-(aziridin-1-yl)-2,4-dinitrobenzamide] (compound 1, scheme 1) is reported to be a substrate for the aerobic bacterial nitroreductase NTR (nfsB gene product) isolated from E. coli (Boland et al., Biochem. Pharmacol. 1991, 41, 867-875; Anlezarket al., Biochem. Pharmacol, 15, 1992, 44, 2289-2295; Parkinson et al., J. Med. Chem. 2000, 43, 3624). This compound has been used as a prodrug in both ADEPT (Knox et al., Biochem. Pharmacol., 1995, 49, 1641-1647) and GDEPT (Bridgewater et al., Eur. J. Cancer, 1995, 31A, 2362-2370; Bailey et al., Gene Ther., 1996, 3, 1143-1150; Bailey and Hart, Gene Ther., 1997, 4, 80-81; Green et al., Cancer Gene Ther., 1997, 4, 229-238) applications, including a clinical trial (Chung-Faye et al., Clin. Cancer Res., 2001, 7, 2662-2668). Similarly, the dinitrophenyl mustard SN 23862 (compound 2, scheme 1) is also a substrate for E. coli NfsB, and shows selective toxicity towards cell lines that express the enzyme. It is activated by nitro group reduction (Palmer et al., J. Med. Chem., 1995, 38, 1229; Kestell et al., Cancer Chemother. Pharmacol., 2000, 46, 365-374). The 4-SO2Me derivative (compound 3, scheme 1) was also a substrate (Atwell et al., Anti-Cancer Drug Des., 1996, 11, 553), as was the dibromo mustard analogue (compound 4, scheme 1) (Atwell et al., J. Med. Chem., 2007, 50, 1197-1212). Prodrugs 1-4 (scheme 1) have poor aqueous solubility. For example, to determine the efficacy of prodrug 4 in xenograft-bearing nude mice, it was administered in either neat DMSO or DMSO/polyethylene glycol/water (Atwell et al., J. Med. Chem., 2007, 50, 1197-1212) resulting in a large variations in maximum tolerated dose.

Some phosphate analogues of mustards have been described, for the purpose of solubilising the compounds. The best known is estramustine phosphate, which has been shown to bind to tubulin binding domains on various microtubule-associated proteins (Moraga et al., Biochim. Biophys. Acta, 1992, 1 121, 97-103), and which has been shown to be active in advanced breast cancer (Keren-Rosenberg et al., Semin. Oncol., 1997, 24(Suppl. 3), 26-29), but has not been shown to be activated by NTR.
Dinitrobenzamide mustards bearing alcohol side chains pendant at a carboxamide (—CONH—) group and their phosphate derivatives are described as bioreductive drugs for GDEPT applications (WO/2008/030112 and WO/2005/042471). Central to the disclosure are prodrugs that provide cell ablation with substantially minimal bystander effect, a term used to describe the back diffusion of cytotoxic metabolites from bacterial nitroreductase-expressing target cells to ablate bacterial nitroreductase naive cells. No bystander efficiency data is provided.
The ability to sterilise neighbouring cells otherwise unable to activate the targeted cytotoxic agent is of central importance to the activity of the agents in combination with nitroreductase enzymes. Gene/enzyme delivery technologies utilised in approaches such as GDEPT, VDEPT, CDEPT and ADEPT are inherently heterogeneous, necessitating efficient redistribution of activated cytotoxic metabolites to inhibit a larger population of neighbouring cells. Thus the bystander effect is an important mechanism to compensate for this anticipated heterogeneity by generating cytotoxic metabolites that diffuse locally to ablate neighbouring vector-naïve cells.
In addition to activation by exogenous oxygen-independent two-electron nitroreductases it is desirable to design nitroaromatic prodrugs, bearing a nitro substituent of an appropriate electron affinity that it is able to be reduced by endogenous human one-electron reductases to produce a nitro radical anion that can be readily back-oxidised by molecular oxygen. In well-oxygenated tissues in the body the parent prodrug is re-formed in a futile redox cycle, however in the presence of pathological hypoxia found in human solid tumours, net reduction to hydroxylamine and amine cytotoxic metabolites is able to occur providing tumour-selective cell killing. Such compounds are termed hypoxia-activated prodrugs (HAP) or hypoxia-selective cytotoxins (HSC).
The Phase II clinical candidate PR-104 is a 3,5-dinitrobenzamide water-soluble phosphate pre-prodrug that, following hydrolysis by systemic phosphatases, releases the ‘hypoxia-activated’ and ‘bacterial nitroreductase-activated’ prodrug PR-104A. Metabolism of PR-104A by endogenous human one-electron reductases in hypoxic cells of a tumour or by exogenous oxygen-independent two-electron nitroreductases, such as bacterial nitroreductases genetically engineered to be expressed in a tumour, produces the DNA crosslinking mustard cytotoxic metabolites PR-104H and PR-104M (Scheme 2) (Patterson et al., Clin Can Res 2007, 13:3922-32).

Unexpectedly PR-104A is also subject to 2e-reduction by an endogenous human reductase called aldo-ketoreductase 1C3 (AKR1C3). This aerobic pathway yields identical cytotoxic metabolites. Expression of AKR1C3 in human CD34+ myeloid progenitor cells may result in a lack of selectivity of PR-104 for solid tumours versus normal bone marrow, compromising PR-104's therapeutic index. It is desirable therefore to eliminate this off-mechanism aerobic activation of PR-104 by AKR1C3.
It is an object of the present invention to provide one or more prodrugs that are substantially free of activation by human AKR1C3 enzyme, or at least to provide the public with a useful choice.